论文部分内容阅读
本文回顾了学习者作文自动评分的理论与实践,提出了综合利用基于向量空间模型和依存语法计算句子相似度算法和专业译者译文语料库的学习者译文自动评价方法。该方法以学习者译文和语料库参考译文语句为单位,通过结合语义与句法相似度的综合句子相似度的计算,考察学习者译文与语料库中的专业译者(参考)译文的相似性,从而获得学习者译文的机器综合评分。实验结果表明,本文的方法充分兼顾了学习者译文与参考译文的语义相似度与句法相似度,其自动评分结果与人工评分结果之间具有较高的相关度。
This paper reviews the theory and practice of learners’ automatic grading of essay composition and puts forward an automatic evaluation method of learner’s translation based on vector space model and dependency grammar to calculate sentence similarity algorithm and professional translator corpus. This method takes the learner’s translation and the corpus as reference. The similarity between the translation of the learner’s translator and the professional translator (reference) in the corpus is studied through the calculation of the syntactic similarity between the semantic and the syntactic similarity, so as to obtain the similarity Learner’s translation of the machine comprehensive score. The experimental results show that the proposed method takes into account the semantic similarity and syntactic similarity of the translation between the learner and the reference translation, and has a high correlation between the automatic score and the artificial score.